Exploring the role of the potential surface in the behaviour of early warning signals.

The theory of critical slowing down states that a system displays increasing relaxation times as it approaches a critical transition. These changes can be seen in statistics generated from timeseries data, which can be used as early warning signals of a transition. Such early warning signals would be of value for emerging infectious diseases or to understand when an endemic disease is close to elimination. However, in applications to a variety of epidemiological models there is frequent disagreement with the general theory of critical slowing down, with some indicators performing well on prevalence data but not when applied to incidence data. Furthermore, the alternative theory of critical speeding up predicts contradictory behaviour of early warning signals prior to some stochastic transitions. To investigate the possibility of observing critical speeding up in epidemiological models we characterise the behaviour of common early warning signals in terms of a system's potential surface and noise around a quasi-steady state. We then describe a method to obtain these key features from timeseries data, taking as a case study a version of the SIS model, adapted to demonstrate either critical slowing down or critical speeding up. We show this method accurately reproduces the analytic potential surface and diffusion function, and that these results can be used to determine the behaviour of early warning signals and correctly identify signs of both critical slowing down and critical speeding up.

How to measure post-error slowing: The case of pre-error speeding.

Post-error slowing is one of the most widely employed measures to study cognitive and behavioral consequences of error commission. Several methods have been proposed to quantify the post-error slowing effect, and we discuss two main methods: The traditional method of comparing response times in correct post-error trials to response times of correct trials that follow another correct trial, and a more recent proposal of comparing response times in correct post-error trials to the corresponding correct pre-error trials. Based on thorough re-analyses of two datasets, we argue that the latter method provides an inflated estimate by also capturing the (partially) independent effect of pre-error speeding. We propose two solutions for improving the assessment of human error processing, both of which highlight the importance of distinguishing between initial pre-error speeding and later post-error slowing.

Self-Reported Speeding Among New York City Adult Drivers, 2015-2016.

Motor vehicle crashes are a leading cause of injury related deaths. Urban areas accommodate multiple road users and pedestrians account for a larger share of traffic fatalities. Speed reduction has been one component of New York City's multidisciplinary approach to reduce traffic fatalities-Vision Zero. Data from the New York City (NYC) Community Health Survey 2015-2016 were used to document population-based estimates of self-reported speeding (defined as driving ten miles per hour or more over the posted speed limit in the past 30 days) among NYC adult drivers collected soon after the adoption of Vision Zero in 2014. Self-reported speeding is common, with nearly two-thirds (63%) of adult drivers indicating they ever sped and 13% often speeding. In adjusted multivariable models, often speeding was more common among younger drivers vs. older drivers (adjusted prevalence ratio: 2.77; 95%CI 1.93-3.98), males vs. females (adjusted prevalence ratio: 1.59; 95%CI 1.35-1.87), wealthier drivers vs. poorer drivers (adjusted prevalence ratio: 1.37; 95%CI 1.10-1.70) and those reporting worse perceived social cohesion vs. better perceived social cohesion (adjusted prevalence ratio 1.51; 95%CI 1.09-2.10). Population-based health surveys facilitate exploration of a range of potential influences on health behaviors.

Executive function and dangerous driving behaviors in young drivers.

The purpose of the present study was to investigate the behavioral and cognitive processes underlying dangerous driving behaviors. We used a survey to assess levels of executive function in college students. The sample consisted of 59 males and 77 females and their age ranged from 18 to 24. We stratified the students into two groups based on executive function scores and compared the extent to which each group engaged in four dangerous driving behaviors (texting while driving, driving without a seat belt, driving while intoxicated, and speeding) as well as how often they experienced three negative driving outcomes (crashes, pulled over, and ticketed). We also investigated how these driving behaviors and outcomes are correlated with subcategories of executive function. The results show that students with a low level of executive function were more likely to engage in dangerous driving behaviors and more likely to experience negative driving outcomes. The results also show that texting while driving, driving while intoxicated, and speeding were most strongly correlated with the executive function subcategory of Impulse Control, whereas driving without a seat belt was most strongly correlated with the executive function subcategory of Strategic Planning. These results suggest that different behavioral or cognitive processes are involved in different dangerous driving behaviors and different interventions may be needed to target each underlying process.

Risk indicators for road accident in Poland for the period 2004-2017.

OBJECTIVE: The aim of the research was to assess the prevalence, effects and risk factors for motor vehicle accidents (MVA) in the years 2004-2017 in Poland. METHOD: We merged secondary data from the Prevention and Analysis Office of Road Traffic Headquarters of Police and Central Statistical Office of Poland. RESULTS: Our results suggest that several thousand people are killed on Polish roads annually, and tens of thousands are injured. Road crashes represent the leading cause of death for Polish men up to 44 years of age. The most common causes of road crashes in Poland include failure to comply with the road traffic rules and low driving skills. We also found drivers who poorly assessed road situations, roads characterized by a lack of adequate road infrastructure, and many vehicles in poor condition. Road crashes have become a significant public health and social problem globally. Drivers caused most MVA in Poland in the years 2004-2017, whereas the underlying cause was inadequate speed regarding the road traffic condition as well as not respecting the right-of-way. Despite various measures that are being taken to improve safety on Polish roads, the number of the dead and wounded as a result of road accidents is still high and the losses to the society are considerable as well. CONCLUSION: It is necessary to continue multidirectional actions to improve safety on the roads in Poland resulting in a systematic increase in the level of road traffic security.

Driving and attention deficit hyperactivity disorder.

Adults with attention deficit hyperactivity disorder (ADHD) suffer from various impairments of cognitive, emotional and social functioning, which can have considerable consequences for many areas of daily living. One of those areas is driving a vehicle. Driving is an important activity of everyday life and requires an efficient interplay between multiple cognitive, perceptual, and motor skills. In the present study, a selective review of the literature on driving-related difficulties associated with ADHD is performed, seeking to answer whether individuals with ADHD show increased levels of unsafe driving behaviours, which cognitive (dys)functions of individuals with ADHD are related to driving difficulty, and whether pharmacological treatment significantly improves the driving behaviour of individuals with ADHD. The available research provides convincing evidence that individuals with ADHD have different and more adverse driving outcomes than individuals without the condition. However, it appears that not all individuals with ADHD are affected uniformly. Despite various cognitive functions being related with driving difficulties, these functions do not appear helpful in detecting high risk drivers with ADHD, nor in predicting driving outcomes in individuals with ADHD, since impairments in these functions are defining criteria for the diagnoses of ADHD (e.g., inattention and impulsivity). Pharmacological treatment of ADHD, in particular stimulant drug treatment, appears to be beneficial to the driving difficulties experienced by individuals with ADHD. However, additional research is needed, in particular further studies that address the numerous methodological weaknesses of many of the previous studies.

Prevalence of drink driving and speeding in China: a time series analysis from two cities.

OBJECTIVES: To confront the public health challenge imposed by road traffic injuries in China. STUDY DESIGN: A consortium of international partners designed and implemented targeted interventions, such as social media campaigns, advocacy for legislative change and law enforcement training, to reduce the percentage of drink driving and speeding in two Chinese cities, Dalian and Suzhou, from 2010 to 2014. METHODS: Time series models were developed to detect changes in the prevalence of drink driving and speeding using data collected through four years of observational studies. RESULTS: This analysis, based on 15 rounds of data, shows that from May 2011 to November 2014, the percentage of vehicles driving above the speed limit decreased from 31.8% (95% confidence interval [CI]: 29.2-34.5) to 7.4% (95% CI: 7.0-7.9) in Dalian and from 13.5% (95% CI: 11.7-15.5) to 6.9% (95% CI: 6.4-7.4) in Suzhou. Drink driving decreased from 1.7% (95% CI: 1.1-2.4) in January 2011 to 0.5% (95% CI: 0.2-0.9) in November 2014 in Dalian and from 6.4% (95% CI: 5.4-7.4) to 0.5% (95% CI: 0.1-2.4) in Suzhou during approximately the same period. Time series models confirmed declining trends in both risk factors in both cities (P-value: 0.06 for speeding prevalence in Suzhou; all other P-values are below 0.05). Disaggregated by vehicle type, saloon cars and SUVs were more likely to exceed the posted speed limit than other types of vehicles in both cities. The speeding rate was higher where the posted speed limit is lower. In Dalian, more drivers were driving above the posted speed limit on weekdays than on weekends (11.4% vs 6.8%); Suzhou had a similar pattern, but the difference was smaller (14.0% vs 12.2%). CONCLUSION: Despite the challenge in accurately attributing the observed changes to one programme, the substantial reduction in the prevalence of the two risk factors suggests that through coordinated actions, internationally recognized best practices in road safety may be effective in improving road traffic safety in China.

Evaluation of the five-year Bloomberg Philanthropies Global Road Safety Program in the Russian Federation.

Road traffic injuries are a leading cause of morbidity and mortality in the world. In Russia, a road safety program was implemented in Lipetskaya and Ivanovskaya oblasts (regions) as part of a 10-country effort funded by Bloomberg Philanthropies. The program was focused on increasing seat belt and child restraint use and reducing speeding. The primary goals of this monitoring and evaluation study are to assess trends in seat belt use, child restraint use, and speed compliance in the two oblasts over the 5 years and to explore the overall impact of the program on road traffic injury and death rates. Primary data via roadside observations and interviews, and secondary data from official government sources were collected and analyzed for this study. Our results indicate significant improvements in seat belt wearing and child seat use rates and in prevalence of speeding in both intervention oblasts. The observations were consistent with the results from the roadside interviews. In Lipetskaya, restraint use by all occupants increased from 52.4% (baseline, October 2010) to 77.4% (final round, October 2014) and child restraint use increased from 20.9% to 54.1% during the same period. In Ivanovskaya, restraint use by all occupants increased from 48% (baseline, April 2012) to 88.7% (final round, October 2014) and child restraint use increased from 20.6% to 89.4% during the same period. In Lipetskaya, the overall prevalence of speeding (vehicles driving above speed limit) declined from 47.0% (baseline, July 2011) to 30.4% (final round, October 2014) and a similar pattern was observed in Ivanovskaya where the prevalence of speeding decreased from 54.6% (baseline, March 2012) to 46.6% (final round, October 2014). Through 2010-2014, the road traffic crash and injury rates per 100,000 population decreased in Lipetskaya oblast (191.5 and 246.9 in 2010 and 170.4 and 208.6 in 2014, respectively) and slightly increased in Ivanovskaya oblast (184.4 and 236.0 in 2010 and 186.7 and 243.4 in 2014, respectively). These road safety improvements are associated with the program that enabled a combined focus on policy reform, legislation, enforcement, advocacy, education, and data collection and use. However, the existence of other road safety efforts, lack of data from comparable regions, and unavailability of risk factor-specific data make it difficult to attribute these changes to the program.

Speeding behavior among teenage drivers during the learner and early independent driving stage: A case study approach.

INTRODUCTION: Speed is a primary contributing factor in teenage driver crashes. Yet, there are significant methodological challenges in measuring real-world speeding behavior. METHOD: This case study approach analyzed naturalistic driving data for six teenage drivers in a longitudinal study that spanned the learner and early independent driving stages of licensure in Maryland, United States. Trip duration, travel speed and length were recorded using global position system (GPS) data. These were merged with maps of the Maryland road system, which included posted speed limit (PSL) to determine speeding events in each recorded trip. Speeding was defined as driving at the speed of 10 mph higher than the posted speed limit and lasting longer than 6 s. Using these data, two different speeding measures were developed: (1) Trips with Speeding Episodes, and (2) Verified Speeding Time. Conclusions & Practical Applications: Across both measures, speeding behavior during independent licensure was greater than during the learner period. These measures improved on previous methodologies by using PSL information and eliminating the need for mapping software. This approach can be scaled for use in larger samples and has the potential to advance understanding about the trajectory of speeding behaviors among novice teenage drivers.

Speeding behavior and associated factors in Buenos Aires, Argentina.

OBJECTIVE: Excessive speed is a major risk factor for serious injuries and death. However, speeding remains a pervasive problem around the world. This study aimed to investigate the factors associated with speeding behavior in the city of Buenos Aires, Argentina. METHODS: A sample of vehicles (n = 34,967) from ten locations in the city was observed in two waves during 2021. Measurements were made at different times and days of the week. Observation sites were free of intersections, traffic lights, speed bumps and cameras, allowing drivers to speed freely. Data on speed, drivers and vehicle types were collected. Factors associated with speeding were identified through logistic regression analyses. RESULTS: 15.3% of vehicles were observed to be speeding. Roads with posted speed limits of 40 km/h showed higher speeding compared to 60 km/h roads. 77% of vehicles traveled above 30 km/h on local roads, and 30% above 50 km/h on avenues. Motorcycles, both commercial and private, showed a higher percentage of speeding compared to all other vehicles. Speeding was lower among women, among adults over 60 years of age, and among those using cell phones. CONCLUSION: It is crucial to strengthen strategies for increased compliance with speed limits. Actions targeting motorcyclists must be a priority.

A comparison of actual and perceived risk of apprehension for speeding in Norway.

This paper compares actual and perceived risk of apprehension for speeding in Norway. Actual risk of apprehension was estimated by relying on data on the number of citations for speeding and the percentage of vehicles speeding when passing automatic traffic counting stations. It was defined as the number of detected violations per million kilometres driven while committing a violation. Perceived risk of apprehension was estimated as a mean annual frequency of getting detected by the police, based on survey answers given by samples of drivers surveyed in 2010, 2014 and 2024. Actual risk of apprehension was converted into a mean annual frequency of detection by relying on estimates of the mean annual driving distance. Thus, perceived mean annual frequency of detection could be compared to actual mean annual frequency of detection. Drivers were found to overestimate the risk of apprehension considerably, but the size of the overestimation declined from 2010 to 2014 and further again to 2024. In 2024, mean perceived risk of apprehension was about 2.4 times higher than actual risk of apprehension. Drivers were also found to overestimate the number of speed cameras deployed in Norway. Only a small minority of drivers had a correct perception of how the risk of apprehension for speeding varied according to the level of speeding. The decisions drivers make about speeding are based on their perceived risk of apprehension; hence it is advantageous to compliance that drivers overestimate the risk of apprehension.

Analyzing the heterogenous effects of factors on high-range speeding likelihood of taxi speeders: Does explainable deep learning provides more insights than random parameter approach?

The random parameters Generalized Linear Model (GLM) is frequently used to model speeding characteristics and capture the heterogenous effects of factors. However, this statistical approach is seldom employed for prediction and generalization due to the challenge of transferring its predefined errors. Recently, the emergence of explainable AI techniques has illuminated a new path for analyzing factors associated with risky driving behaviors. Despite this, there remains a gap that comparing results from machine and deep learning (ML/DL) approaches with those from random parameters GLM. This study aims to apply the random parameter GLM and explainable deep learning to evaluate the heterogenous effects of factors on the taxis' high-range speeding likelihood. Initially, a Beta GLM with random parameters (BGLM-RP) is developed to model the high-range speeding likelihood among taxi drivers. Additionally, XGBoost, a simple convolutional neural network (Simple-CNN), a deeper CNN (DCNN), and a deeper CNN with self-attention (DCNN-SA) are developed. The quantified explanations and illustrations of the factors' heterogenous effects from ML/DL models are derived from pseudo coefficients by decomposing factors' SHapley Additive exPlanations (SHAP) values. All the developed statistical, ML, and DL models are compared in terms of mean absolute errors and mean square errors on testing and full data. Results show that DCNN-SA excels in prediction on testing data, indicating its superior generalization capabilities, while BGLM-RP outperforms other models on full data. The DCNN-SA can reveal the heterogenous effects of factors for both in-sample and out-of-sample data, which is not possible for the random parameter GLM. However, BGLM-RP can reveal larger magnitudes of the factors' heterogenous effects for in-sample data. The signs and significances are identical between the varying coefficients from BGLM-RP and the pseudo coefficients from the ML/DL models, demonstrating the validity and rationale of using the proposed explanation framework to quantify the factors' effects in ML/DL models. The study also discusses the contributions of various factors to the high-range speeding likelihood of taxi drivers.

Naturalistic driving analysis of situational, behavioral, and psychosocial determinants of speeding.

The present analysis used full-trip naturalistic driving data along with driver behavioral and psychosocial surveys to understand the individual and contextual predictors of speeding. The data were collected over a three-week period from 44 drivers and contain 3,798 full trips, with drivers speeding 7.8 % of the time. Speeding events were identified as periods when participants traveled at a velocity greater than five mph over the speed limit for at least five seconds. Data were analyzed using the Comprehensive Driver Profile (CDP) framework which uses principal component analysis (dimensionality reduction), random forest (predictive modeling), k-means clustering (grouping and profiling), and bootstrapping (profile stability) to decompose the predictive variables and driver characteristics. The final dataset included 188 candidate independent variables from the CDP framework and one dependent variable (speeding). Nine variables emerged as significant predictors of speeding onset with an AUC of 0.88, including the percent of trip time spent idling and speeding, highway driving in low traffic conditions, and positive attitudes toward phone use. Percent of trip speeding was associated with a higher likelihood of speeding by up to 42 percent, and percent trip idling was associated with it by up to 30 percent. Driver profile clusters revealed four types: Traffic & Idling Speeders, Infrequent Speeders, Frequent Speeders, and Situational Speeders. The present analysis demonstrates the importance of situational factors and individual differences in motivating speeding behavior. Countermeasures targeting speeding may be more effective if they address the root causes of the behavior in addition to the behavior itself.

Post-loss speeding or post-win slowing? An empirical note on the interpretation of decision-making time as a function of previous outcome.

Differences in response time following previous losses relative to previous wins are robust observations in behavioural science, often attributed to an increased (or decreased) degree of cognitive control exerted after negative feedback, hence, post-loss slowing (or post-loss speeding). This presumes that the locus of this effect resides in the specific modulation of decision time following negative outcomes. Across two experiments, I demonstrate how the use of absolute rather than relative processing speeds, and the sensitivity of processing speeds in response to specific experimental manipulations (Experiment 1: win rate, Experiment 2: feedback), provide clarity as to the relative weighting of post-win and post-loss states in determining these behavioural effects. Both experiments show that the speeding or slowing of decision-time is largely due to the flexibility generated by post-win cognitive states. Given that post-loss speeding may actually represent post-win slowing, conclusions regarding the modulation of decision-making time as a function of previous outcomes need to be more carefully considered.

Analyzing fatal crash patterns of recidivist drivers across genders and age Groups: A hazard-based duration approach.

Identifying factors that significantly affect drivers that are repeatedly involved in traffic violations or non-fatal crashes (defined here as recidivist drivers) is very important in highway safety studies. This study sought to understand the relationship between a set of variables related to previous driving violations and the duration between a previous non-fatal crash and a subsequent fatal crash, taking into account the age and gender of the driver. By identifying the characteristics of this unique driver population and the factors that influence the duration between their crash events strategies can be put in place to prevent the occurrence of future and potentially fatal crashes. To do this, a five-year (2015-2019) historical fatal crash data from the United States was used for this study. Out of 15,956 fatal crashes involving recidivist drivers obtained, preliminary analysis revealed an overrepresentation of males (about 75%). It was also found that the average duration between the two crash events was about a year and a half, with only an average of one month difference between male and female drivers. Using hazard-based duration models, factors such as number of previous crashes, previous traffic violations, primary contributing factors and some driver demographic characteristics were found to significantly be associated with the duration between the two crash events. The duration between the two events increased with driver's age for drivers who were involved in only one previous crash and the duration was shorter for those that were previously involved in multiple crashes. Previous DUI violations, license suspensions, and previous speeding violations were found to be associated with shorter durations, at varying degrees depending on the driver's age and gender. The duration was also observed to be longer if the fatal crash involved alcohol or drug use among younger drivers but shorter among middle-aged male drivers. These findings reveal interesting dynamics that may be linked to recidivist tendencies among some drivers involved in fatal crashes. The factors identified from this study could help identify crash countermeasures and programs that will help to reform such driver behaviors.

Assessing driver behavior in work zones: A discretized duration approach to predict speeding.

Higher speeds in work zones have been linked to an increased likelihood of crashes and more severe crash outcomes. To enhance safety, speed limits are often reduced in work zones, aiming to create a steady flow of traffic and safer traffic operations such as merging and flagging. However, this speed reduction can also lead to abrupt speed changes, resulting from sudden braking or acceleration, increasing the risk of crashes. This disruption in speed and flow results increases the likelihood of rear-end crashes. Ensuring driver compliance with the reduced speed limits and traffic flow operations is challenging as work zones may cause frustration and lead to more instances of speeding. Therefore, proactively predicting speeding events in work zones can be crucial for the safety of both workers and road users, as it enables the implementation of speed enforcement measures to maintain and improve driver compliance in advance. In this study, we employ the duration-based prediction framework to forecast speeding occurrences in work zones. The model is used to identify significant predictors of speeding including visibility, number of lanes, posted speed limit, segment length, coefficient of variation in speed, and travel time index. Among these variables, the number of lanes, posted speed limit, and coefficient of variation of speed are positively associated with speeding. On the other hand, visibility, segment length, and travel time index are negatively associated with speeding. Results show the model's predictive accuracy is higher for speeding events with shorter durations between consecutive occurrences. The model predicted speeding within 61% of the actual epoch when speeding events within 5 h of one another were considered for validation. This indicates that the model is more effective for road segments and work zones where speeding occurs more frequently. The prediction framework can be a great asset for agencies to improve work zone safety in real-time by enabling them to proactively implement effective work zone enforcement measures to control speeding and to stay prepared, preventing potential hazards.

Understanding speeding behavior on interstate horizontal curves and ramps using networkwide probe data.

INTRODUCTION: Lane departure collisions account for many roadway fatalities across the United States. Many of these crashes occur on horizontal curves or ramps and are due to speeding. This research investigates factors that impact the odds of speeding on Interstate horizontal curves and ramps. METHOD: We collected and combined two unique sources of data. The first database involves comprehensive curve and ramp characteristics collected by an automatic road analyzer (ARAN) vehicle; the second database includes volume, average speed, and speed distribution gathered from probe data provided by StreetLight Insight®. We evaluated the impacts of level of service (LOS), which reflects traffic density or level of congestion, time of the day (morning, evening, and off-peak hours), time of the week (weekdays and weekends), and month of the year (Jan-Dec), and various information about geometric characteristics, such as curve radius, arc angle, and superelevation, on odds of speeding. RESULTS: The results show that the odds of speeding increases at horizontal curves with improved levels of service, as well as those with larger radii and superelevation. The odds of speeding decreases on curves with larger arc angles and during the winter months of the year. The findings indicate a reduction in odds of speeding at diagonal/loop ramps with larger arc angles and narrower lane widths. CONCLUSION: The results show the importance of using speed enforcement and other countermeasures to reduce speeding on curves with low traffic volumes, high speed limits, and large radius and superelevation, especially for those in rural areas. PRACTICAL APPLICATION: The results could be used to prioritize locations for the installation of speed countermeasures or dispatch enforcement resources to high-priority locations and times.

Identification of the factors influencing speeding behaviour of food delivery e-bikers in China with the naturalistic cycling data.

With the rapid growth of the gig economy in China, millions of food delivery e-bikers are making their living by rushing on the street. Speeding is one of their most common risky riding behaviours, leading to severe traffic crashes. Based on 2-month naturalistic cycling data of 46 full-time food delivery e-bikers in Changsha, their speeding behaviour is deeply studied with the individual daily speeding proportion being taken as the speeding indicator. A beta regression model is built to identify the factors significantly influencing the indicator. The estimation results reveal that female riders, middle-aged riders and riders with a bachelor's degree are less likely to engage in speeding. The same result is indicated for those working longer or experiencing more crashes. Additionally, holidays and riding distance are found to have significantly positive influences. Finally, some countermeasures are proposed to prevent speeding among food delivery e-bikers.

Improvement of riding licensing criteria based on rider's braking capability: A preliminary study.

Speeding is the factor that usually associated with fatal accident. However, riders have tendency to exceed their vehicle's speed above the regulated speed. Therefore, the likelihood of traffic accidents is significantly influenced by braking ability. Unfortunately, the braking capability has not been accommodated properly in the accident risk management, such as riding license obtaining mechanism. This paper focuses on the possibility of the development of riding licensing criteria based on rider's braking capability. The parameters used in the analysis are the safety factor and margin of safety, due to the differences in riders' braking capability. All the input data were collected from the result of previous related studies. Although the sample size is varied but data source was taken from relevant objects studies. The result of this study showed that impact speed and/or rider's involvement in fatal crashes could be reduced by increasing their braking ability. It strongly indicates that each rider should realize that their speed choices should be suited to their braking ability which could be increased during the riding licensing practical test. The utilization of a rider's braking abilities, which could provide a minimum margin of safety, should therefore be taken into consideration as a basis for the criteria used to obtain a riding license.

Interactive effects analysis of road, traffic, and weather characteristics on shared e-bike speeding risk: A data-driven approach.

As electric bikes (e-bikes) rapidly develop in China, their traffic safety issues are becoming increasingly prominent. Accurately detecting risky riding behaviors and conducting mechanism analysis on the multiple risk factors are crucial in formulating and implementing precise management policies. The emergence of shared e-bikes and the advancements in interpretable machine learning present new opportunities for accurately analyzing the determinants of risky riding behaviors. The primary objective of this study is to examine and analyze the risk factors related to speeding behavior to aid urban management agencies in crafting necessary management policies. This study utilizes a large-scale dataset of shared e-bike trajectory data to establish a framework for detecting speeding behavior. Subsequently, the extreme gradient boosting (XGBoost) model is employed to identify the level of speeding risk by leveraging its excellent identification ability. Moreover, based on measuring the degree of interaction among road, traffic, and weather characteristics, the investigation of the complex interactive effects of these risk factors on high-risk speeding is conducted using bivariate partial dependence plots (PDP) by its superior parsing ability. Feature importance analysis results indicate that the top five ranked variables that significantly affect the identified results of speed risk levels are land use density, rainfall, road level, curbside parking density, and bike lane width. The interaction analysis results indicate that higher levels of road and bike lane width correspond to an increased possibility of high-risk speeding among riders. Land use density, curbside parking density, and rainfall display a nonlinear effect on high-risk speeding. Introducing road level, bike lane width, and time interval could change the patterns of nonlinear effects in land use density, curbside parking density, and rainfall. Finally, several policy recommendations are proposed to improve e-bike traffic safety by utilizing the extracted feature values associated with a higher probability of high-risk speeding.